wavlm-libri-clean-100h-large
This model is a fine-tuned version of microsoft/wavlm-large on the AHAZEEMI/LIBRISPEECH10H - CLEAN dataset. It achieves the following results on the evaluation set:
- Loss: 0.0893
- Wer: 0.0655
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 2
- eval_batch_size: 2
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 4
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 3.0
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 0.0144 | 0.42 | 300 | 0.0947 | 0.0749 |
| 0.1408 | 0.84 | 600 | 0.1347 | 0.1363 |
| 0.0396 | 1.26 | 900 | 0.1090 | 0.0935 |
| 0.0353 | 1.68 | 1200 | 0.1032 | 0.0832 |
| 0.051 | 2.1 | 1500 | 0.0969 | 0.0774 |
| 0.0254 | 2.52 | 1800 | 0.0930 | 0.0715 |
| 0.0579 | 2.94 | 2100 | 0.0894 | 0.0660 |
Framework versions
- Transformers 4.26.1
- Pytorch 2.0.0+cpu
- Datasets 2.9.0
- Tokenizers 0.13.2